Results of this systematic review and meta-analysis were reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2009 guidelines (Supplementary Material) [14]. We systematically searched PubMed and Scopus to identify studies on the prevalence of IA in patients with aortopathy or the presence of aortic disease in people who suffered from IAs published before August 2019. A combination of keywords was used to grab eligible studies: aorta, aortic dissection, aortic aneurysm, aortic dilatation, CoA, BAV aortopathy, and IA. Searching terms were listed in the Supplementary Material. We did not set any other limitation in this search strategy. References of included studies were manually checked to prevent the omission of eligible research.

Two collaborators (X.Y. and L.X.) individually screened the studies from two databases for eligibility according to predefined selection criteria: cohort study, case-control study, and crosssectional study reported an accurate prevalence of IA in patients with aortic disease; indicated a prevalence of aortic disease in patients suffered from IA. In studies evaluating the prevalence of IA, study population should be patients with brain imaging to avoid the omission of cases. Titles and abstracts from database research were examined and ineligible studies were refused. Full texts of remaining publications were carefully reviewed subsequently. We excluded studies on a specific population (e.g., children). Disagreements were solved in a discussion with a senior author (S.C.).

Two investigators (X.Y. and L.X.) independently extracted data on authors, published year, study country, study design (cohort, case-control, or cross-sectional), event (IA or aortic aneurysm), diagnostic criteria (for IA and aortic diseases), number of events, number of participants from included articles. To investigate the effect of cardiovascular risk determinants on the overall prevalence, risk factors (e.g., gender, smoking) in each study were documented in details. We recorded age and the exact number of patients with or without dichotomous risk factor in two groups (patients with both aortopathy and IA; patients with aortopathy alone). In a previous study [8], a differing anatomical distribution pattern of IA between subgroups of the aortic aneurysm was reported, and we tended to find whether this site-specific phenomenon exists between different kinds of aortic disease. We extracted the accurate location and size of IAs provided by original work. If prevalence from reports were adjusted by confounding factors, we grabbed primary data to reduce the heterogeneity between studies. Multiple reports in a single article were analyzed separately. A consensus could be reached through group discussion. If the definite number of events/participants was not provided in eligible studies, we tried to contact authors for more information.

We included BAV, CoA, aortic aneurysm, and aortic dissection as aortopathy (aortic disease) in this literature. BAV was an inherited form of heart disease, but there remained a notable association between BAV and aorta, and we considered BAV an aortic disease in our research. To analyze the site-specific distribution pattern of IA in patients with different aortopathy, IAs were divided into three groups according to their anatomical locations: intracranial internal carotid artery (ICA), anterior choroidal artery, superior hypophyseal artery, and ophthalmic artery (ICA-IA); anterior circulation excluding the ICA, comprising anterior cerebral, anterior communicating, and middle cerebral arteries (anterior circulation-IA [Ant-IA]); and posterior circulation consisted of arteries not included in the ICA-IA or Ant-IA group (Post-IA) [8]. This classification for IAs was originally used by Shin et al. [8]

The methodological quality of included studies (cross-sectional studies) was evaluated using a modified Newcastle-Ottawa scale developed by Fralick et al. [15] The modified scale covered three perspectives (participant selection; comparability; assessment of outcome) of methodology. Two investigators (X.Y. and L.X.) independently assessed the methodological quality. All results were reviewed by a third investigator (S.C.) with disagreements being resolved through consensus. We did not set any eligibility restrictions on the quality assessment score for the studies that fulfilled inclusion criteria.

Three studies reported a prevalence of aortic disease in IA population, and we qualitatively described the result on this issue. The main outcome was a pooled prevalence of IA in patients with aortopathy. The meta-analysis for overall prevalence was conducted in Stata version 14.0 (StataCorp., College Station, TX, USA) and forest plot was prepared in R version 3.6.1 using the forestplot packages (R Foundation for Statistical Computing, Vienna, Austria). Study heterogeneity was fully assessed by the I² statistic (0% to 100%) describing the proportion of total variation. Little heterogeneity was defined as I²<25%, moderate heterogeneity was 25%<I²<50%, and I²>50% stands for a substantial heterogeneity. We expected heterogeneity in the methodology of included studies, hence the meta-analysis was based on a random-effect model. We used the data of participants/events extracted from original articles to calculate prevalence and corresponding 95% confidence interval (CI). Meta-analysis was performed in predefined subgroups of aortopathy (BAV, CoA, aortic aneurysm, and aortic dissection). Publication bias was not detected for the limited number of studies in each subgroup. No sensitivity analysis was planned. We investigated risk factors that may influence the estimated prevalence of IA in patients suffered from aortopathy. All risk factors reported in at least three studies were included in pooled analyses using a random-effect model. Data of dichotomous risk factors were arranged in fourfold tables and odds ratios (ORs) were obtained through calculation. Continuous risk factors were raised as standardized mean differences by Stata version 14.0 and transformed to ORs according to the formula developed in the Cochrane handbook version 5.1.0 [16]. We calculated the OR of risk factors on the platform of IBM SPSS Statistics version 24 (IBM Co., Armonk, NY, USA). Primary ORs were pooled on the platform of Stata 14.0, and the final result was displayed in a forest plot produced by R. Continuous variables were compared in analyses of variance or unpaired Student’s t-test. All P-values were two tails, and P-values <0.05 were considered significant.

The present work is a systematic review with meta-analyses of published studies, and data were collected from published materials, and thus ethical approval was not necessary for this article.

All extracted data could be made available upon request from qualified investigators to corresponding authors.

Figure 1 showed the flow diagram for search strategy and selection procedure of this systematic review. The primary database search yielded 6,982 articles. After removal of duplicates, 6,301 researches were identified and their titles/abstracts were manually retrieved for further evaluation. Of 100 studies considered for eligibility, 13 records were included in the full-text assessment. In the process of screening, one study was conducted in children (mean age 16) with CoA [17], and we excluded this article for the specific study population, and we also excluded a cohort study in which not all participants were patients with brain imaging [18]. After an inter-observer agreement between reviewers for study inclusion, three studies considering the prevalence of aortic disease in patients with IA were included in qualitative analysis, and a total of ten studies were included in the meta-analysis which reported an estimated prevalence of IA among patients with aortopathy.

Overall, 13 cross-sectional studies were included in this systematic review (Table 1). Three studies investigated the proportion of patients with aortic aneurysm or BAV in patients suffered from IA [19-21]. Ten studies reported a prevalence of IA in patients with aortopathy: two studies were in patients with BAV [10,13]; three studies considered patients with CoA [9,11,12]; three studies were conducted in patients with aortic aneurysm [5,7,8]; one study investigated aortic dissection [3]; and one study examined both aortic aneurysm and aortic dissection [6]. Seven studies were from the USA, and three studies from South Korea. One study was from Japan, and one study from the UK, and one study from Finland. Of the total 4,041 participants, 1,261 patients were suffered from IA and 3,193 were patients struggled with the aortic disease. IAs were diagnosed by computed tomography angiography, magnetic resonance angiography (MRA) or digital subtraction angiography, and aortic diseases were identified by echocardiography, computed tomography, and MRA. Quality assessment of included studies was displayed in Supplementary Table 1, and overall study quality was moderate (6 to 7 points). We did not exclude any study on the basis of the quality assessment.

Of three included studies, Goyal et al. [19] indicated that the prevalence of BAV and thoracic aortic aneurysm in IA patients were 0.6%, 4.7% respectively. Miyazawa et al. [20] reported a prevalence of 7.2% of abdominal aortic aneurysm in patients with IA. Laukka et al. [21] showed a higher prevalence of thoracic aortic aneurysm (7.5%) in IA patients compared to the general population. Three studies suggested several risk factors (e.g., age and smoking) for this association, but we could not further analyze these factors due to the limited data.

There are several mechanisms for the coexistence of IAs and aortic diseases. The abnormality of cells derived from the neural crest could explain the association between IA and aortopathy [10]. The neural crest cells are a group of transient and multipotent cells in early embryogenesis which is induced to migrate and give rise to several cell lineages. The tunica media of the aortic arch and its branches, including the cervicocephalic arteries are composed of cells derived from neural crest, and the abnormal development of the neural crest cells may result in vascular fragility, and this could cause simultaneous IAs and aortic diseases [10,22-25].

Furthermore, the genetic relation of IA and aortopathy has been examined in researches [26-34]. Multiple types of gene mutation play a critical role in this relationship: single nucleotide polymorphisms [27,35], frameshift mutations in exons [28], and translocations of chromosome [29], which demonstrates that the gene-related association between IA and aortopathy is rather complicated.

In particular, patients with connective tissue diseases such as Loeys-Dietz syndrome, Marfan syndrome, Ehlers-Danlos syndrome, and neurofibromatosis type 1 are thought to be at a higher risk for IAs and aortopathy [36,37]. The widely accepted causal explanation of this relationship is that the genetic mutations in such connective tissue disorders affect the collagen and proteoglycans in extracellular matrix, thus leading to the weakening of the vascular wall [37].

Acquired risk factors could also increase the risk of IA and aortic disorders. The dominant environmental risk factors for aneurysms include age, gender, smoking, alcohol, and hypertension [35,38]. Hypertension is a known risk factor for the development of IA [35], aortic aneurysm [39], and aortic dissection [40]. Blood flow or blood pressure is an important modulator of arterial growth. Wall shear stress is strictly regulated in bodies, with the normal artery enlarging in response to increases in blood flow to maintain normal shear stress [35]. However, high blood pressure in patients with hypertension could significantly affect the shear stress, which could lead to an abnormal expansion of blood vessels.

The pooled prevalence of IA among patients with distinct aortic disease in this research is similar to that of previous observational study [3,5-13]. The estimated prevalence of IA for patients with aortic dissection (23%; 95% CI, 12% to 34%) is much higher compared with that in groups suffered from other aortic diseases. Study quality is moderate and heterogeneity source could be identified through subgroup analysis. The heterogeneity within or between subgroups is limited, which indicates that the relationship between IA and aortopathy is robust regardless of the type of aortopathy. The overall estimated prevalence of IA is around 12% in patients with aortopathy, which is quadrupled compared to that of the general population, and the prevalence of IA in patients suffered from BAV, CoA, aortic aneurysm, aortic dissection is 8%, 10%, 12%, and 23% respectively. BAV and CoA are congenital diseases, and aortic aneurysm or dissection could be induced by acquired factors, and the prevalence of aortic aneurysm or dissection is higher, which indicates that hemodynamic or environmental factors play an essential role in the pathogenesis of IA and aortopathy. The result emphasizes that IA screenings could be conducted among patients with aortic diseases, especially patients with a dissection of the aorta. Importantly, the cost for population screening for IA is considerable, and a cost-effective study in this field is required.

In the surgical intervention of aortopathy, patients could suddenly die for a nubilous reason, and this might be associated with IA or other cerebrovascular accidents. Our findings should be relevant for neurologists, cardiac surgeons, patients with aortopathy and policymakers.

In this research, the effect of gender, age, smoking, hypertension, diabetes, hyperlipidemia, and the overlapping with two aortic diseases on the prevalence of IA in aortopathy sufferers are quantitatively analyzed (Figure 3). Gender (female) and smoking could be risk factors for IAs among patients with aortic diseases. However, study numbers and sample sizes in pooled analyses for risk factors are small (Figure 3), and it could somewhat reduce the statistical power of analyses. Thus, large observational studies are needed to identify precise risk factors for IAs among patients with aortopathy. Priority should be given to patients with special risk factors in population screening for IAs.

A site-specific distribution pattern of IA between groups with different aortopathy is observed in the review. The anatomical distribution of IAs differed significantly in patients with distinct aortic disease. Despite a heterogeneous frequency of ICA-IA and Post-IA among different groups, the frequency of Ant-IA is confoundedly stable among patients suffered from distinct aortopathy. Additionally, the frequency of Ant-IA is the largest in three predefined IA groups. Genetic risk factors are suggested to have a larger role of IA at the middle cerebral artery (contained in Ant-IA) than at other sites of the cerebrovascular system [41]. Consistently, genetic factors are critical in the development of BAV, CoA, aortic aneurysm, and aortic dissection [42-45]. The result suggests a shared pathological mechanism between IA and aortopathy [8]. The stable distribution of Ant-IA might suggest that the connection between aortopathy and IA is strongly associated with the human gene. Distinct allocation of IAs in ICA-IA and Post-IA could accompany with exclusive pathological features of aortic diseases, and this issue should be further explored. We expect subtypes of aortic aneurysm, and we analyze the site of IA in thoracic aortic aneurysm and abdominal aortic aneurysm separately. However, the margin between two groups was not statistically significant, which is inconsistent with a previous study [8]. Shin et al. [8] deliver a differing distribution patterns of IA among subgroups of aortic aneurysm (ascending aneurysm, descending thoracic-suprarenal aneurysms, and infrarenal aneurysm) with 71 participants. We consider the study population is too small to confirm the relationship, and different classifications could somewhat affect the result. It is known that genes in the pathogenesis of IA and thoracic aortic aneurysm are closely related, and the pathogenesis of abdominal aortic aneurysm is tended to be more atherosclerotic, but both thoracic aortic aneurysm and abdominal aortic aneurysm affected by hemodynamic and environmental factors. This could explain why the site distribution of IA in two groups is not totally different. Since the sample size of the site-specific analysis is still small, future work is needed on this issue to further test the relationship.

The qualitative part of this review analyzes the presence of aortic disease in patients with IA. Reports from included studies [18-20] lack credible controls that could provide a reliable overall incidence/prevalence of aortic disease in the general population. Unlike IAs, aortic diseases get considerable symptoms and multiple diagnostic methods, and usually it is easier to diagnose aortic disease than IA. However, as increasing small IAs are found incidentally on neuroimaging, a substantial group of patients with IAs may benefit from screening for aortic vascular pathology.

To the best of our knowledge, this study is the first comprehensive systematic review of the relationship between IA and aortopathy. We focus on the impact of cardiovascular risk factors on the overall prevalence, and the result indicates that the presence of aortopathy could be a predictor of IA. The specific distribution pattern of IAs in distinct aortic disease is also mapped in this research. The current review has several limitations. The methodological differences (e.g., diagnostic methods for events) in the original studies lead to inevitable heterogeneity. Although we include 10 studies to quantify the prevalence and the effect of risk factors on the prevalence, the small sample size could somewhat reduce the statistical power and increase the error of statistical analyses. Other risk factors (e.g., alcohol, history of stroke) are also reported, but we are incapable of evaluating these factors due to the limited studies. Complete records of the number and anatomical location for IA in single participants are not fully provided, we are unable to assess the association between different aortopathy and the number of IAs in individuals.